New Żywiec 0.0% campaign with Machine Learning

case studies

published

For the Żywiec 0.0% brand campaign, we have created a website with competition based on voice recognition. We used the Machine Learning algorithm that recognizes the voice and awards the participants with appropriate prizes for the precisely sung phrase of the song.

task

01

The Żywiec 0.0% brand in its latest campaign has planned a competition consisting in singing a theme from the spot - "TURURU" to the beat of the song "The Final Countdown". An algorithm that analyzes the precision of the competition task decides about winning. Depending on the accuracy of singing the phrase "turururu", the contestant gets the appropriate prize - the winners receive e-mail codes entitling them to receive Żywiec 0.0% beers or purchase non-alcoholic beer variants with a 50% discount.

solution

02

As part of the preparations, we created a prototype enabling the collection of singing samples from the test group. They were used to feed the algorithm based on machine learning and, as a result, teach it to recognize the phrase "turururu". Alexa and Siri work in a similar way, they respond to specific commands. After the learning stage, the final version of the algorithm was plugged into the Żywiec 0.0% competition website.

 

To improve the performance of the contest application, we used on-the-fly voice processing technology, without sending it to the server. This makes participants immediately know if they are singing rhythmically and can get one of the prizes. The correctness of the task performance can be checked on the animation of a filling bottle of Żywiec 0.0% - when the singers hit the appropriate sounds, the level of beer increases.


We created the website in cooperation with agencies: TWIN .digital collective and Feeders Agency.

The solutions used by us ensure that the participant immediately receives feedback - the filling bottle constantly signals the accuracy of the competition task.